CN107104899B - Ant colony algorithm-based routing method applied to vehicle-mounted self-organizing network - Google Patents

Abstract

The invention provides an ant colony algorithm-based routing method applied to a vehicle-mounted self-organizing network, which comprises the following steps: the Q learning process is completed in the initial stage of network operation by the vehicle nodes in the VANET through mutual transmission and the process of updating a Q value table; the route discovery process is a process of finding an optimal route as required; the discovered routes rely on route maintenance procedures to achieve maintenance. The method is used for route discovery and route maintenance, establishes an updating mechanism of pheromones, applies a standard Q learning principle, accelerates the convergence of the algorithm to adapt to VANET network topology with high change frequency, and simultaneously obtains a link duration reference value by combining a relevant mobile prediction method to ensure the link quality and enhance the path effectiveness. The method has better adaptability to the VANET environment, fully considers the duration of the link to ensure the link quality during the route selection, accelerates the route establishment process on the premise of ensuring the link quality, can quickly react to the dynamic change of the network, and ensures the effectiveness of data communication.

Description

Ant colony algorithm-based routing method applied to vehicle-mounted self-organizing network

Technical Field

The invention relates to the technical field of vehicle-mounted self-organizing routing protocols and vehicle-mounted self-organizing networks, in particular to a routing method based on an ant colony algorithm and applied to the vehicle-mounted self-organizing networks.

Background

The vehicle-mounted ad hoc network is an important component of an intelligent traffic system and is a special wireless ad hoc multi-hop network. The VANET provides network access service for passengers, drivers and traffic managers, and establishes a temporary network of self-organization distributed control on a highway network quickly and dynamically by means of direct or indirect multi-hop communication between vehicles, vehicles and pedestrians and between vehicles and roadside base stations. In the VANET, a vehicle carrying wireless communication equipment is used as a network node, and the vehicle node provides information receiving, transmitting and relay forwarding functions for a user.

The ant Colony optimization algorithm is firstly proposed by m.Dorigo through foraging research of ant Colony in nature, and is optimized to provide an ant Colony optimization algorithm ACO (Ant Colony optimization). Subsequently Holland, rotkrantz, Bruten and Schoonderwoerd were the earliest to apply the ACO algorithm to network routing and the ABC (anti-based control) algorithm was proposed. With the rise of the vehicle-mounted ad hoc network, the ant colony algorithm draws attention of researchers to the strong adaptability of the vehicle-mounted ad hoc network.

Disclosure of Invention

The invention provides an ant colony algorithm-based routing method applied to a vehicle-mounted self-organizing network for overcoming at least one defect (deficiency) in the prior art, the method applies the ant colony algorithm to the research of a VANET routing protocol, has better adaptability to the VANET environment, and can quickly respond to the dynamic change of the network on the premise of ensuring the link quality. The packet delivery rate of the Q-PABRP is superior to that of the AODV protocol, and the Q-PABRP has higher transmission reliability compared with the AODV protocol.

In order to achieve the purpose, the technical scheme of the invention is as follows:

the ant colony algorithm-based routing method is applied to the vehicle-mounted self-organizing network, and specifically comprises the following steps:

firstly, the method comprises the following steps: the Q learning process is implemented by the maintenance ant HELLO. In the initial stage of network operation, initializing a Q value table of each node, wherein the initial value is 0, exchanging information in the Q value table by sending a HELLO packet to a neighbor node by each node, and reading the Q value information to update the Q value table of each node after the node receives the HELLO packet. In the initial phase of network operation, each node broadcasts a HELLO packet to neighboring nodes. After receiving the HELLO packet from the neighbor node, the current node judges whether the current node has the information of the neighbor node which sends the HELLO packet. If the information of the neighbor node does not exist, the neighbor node is added into the routing table of the neighbor node, and the information of the corresponding routing table entry is updated. The node reads the value in the HELLO packet and updates its Q value table. And after the Q value is updated, updating the transition probability information in the routing table entry of the corresponding node. And after the current node finishes the updating work of the information, discarding the HELLO packet.

Secondly, the method comprises the following steps: the route discovery process is realized by forward ant FANT and backward ant BANT together. When a network node wants to initiate data communication, a source node first checks whether routing information to a destination node exists, and if the routing information does not exist, a route discovery process is triggered.

Thirdly, the method comprises the following steps: the route maintenance process is realized by detecting ant EXPLORE and maintaining ant HELLO together, and the aim is to ensure the smooth end-to-end communication. This is achieved depending on the routing duration. When the time for establishing the routing table entry exceeds the value of the routing duration in the corresponding table entry, the node sends an EXPLORE to the destination node. In order to make the operation of the explicit packet more efficient, the state of the neighbor node needs to be confirmed at any time, and this process is implemented by the HELLO packet. When the HELLO packet in the protocol is improved, the HELLO packet is periodically broadcast to the neighbor nodes, and when the HELLO packet sent by a certain neighbor node is not received by the node for a certain time, the neighbor nodes are automatically considered to be absent, the table information related to the node is deleted, and then an EXPLORE packet is sent out to find out a new route of the destination node.

The invention relates to an ant colony algorithm-based routing method applied to a vehicle-mounted self-organizing network, which is characterized in that an ant colony algorithm is applied to a VANET routing protocol research, a routing process is simulated as an ant foraging process, the convergence speed and a link stability mechanism of the routing algorithm are comprehensively considered, and a routing protocol Q-PABRP based on the ant colony algorithm is suitable for the VANET.

The invention applies an Ant colony algorithm to research of a VANET Routing Protocol, simulates a Routing process as a process of Ant foraging, comprehensively considers the convergence speed and the link stability mechanism of the Routing algorithm, and provides a Routing Protocol Q-PABRP (Q-Learning predictive and based Routing Protocol) suitable for the VANET and based on the Ant colony algorithm. Simulation results show that the algorithm has better adaptability to the VANET environment and can quickly respond to the dynamic change of the network on the premise of ensuring the link quality.

Compared with the prior art, the invention has the following advantages and beneficial effects:

by fully considering the advantages of the ant colony algorithm and the applicability of the ant colony algorithm in the VANET environment, the ant colony algorithm is used as a research foundation and an entry point of the VANET route, an ant colony algorithm model in the network route is constructed, Q learning and link quality evaluation are combined, a routing protocol Q-PABRP suitable for the VANET is provided, and route convergence is accelerated on the premise of ensuring link quality.

And performing related performance comparison on the Q-PABRP and the AODV aiming at the performance evaluation index of the routing protocol. The packet delivery rate of the Q-PABRP is superior to that of the AODV protocol because the AODV protocol uses the minimum hop count as a routing criterion and does not consider the influence of reverse traffic, and the Q-PABRP uses the link duration to ensure the validity of the communication link, so that the Q-PABRP has higher transmission reliability compared with the AODV protocol. The Q-PABRP considers the influence of the link duration on the effectiveness problem of the route, and reduces the possibility of the occurrence of the route which is easy to fail, so that the Q-PABRP is superior to the AODV protocol in the performance.

The invention has better adaptability to VANET environment and can quickly respond to the dynamic change of the network on the premise of ensuring the link quality.

Drawings

Fig. 1 is a flow chart of the processing of a HELLO packet.

Fig. 2 is a flow chart of the process for FANT grouping.

Fig. 3 is a flow chart of the process for grouping BANTs.

Fig. 4 is a flow chart of a route maintenance process.

Detailed Description

The drawings are for illustrative purposes only and are not to be construed as limiting the patent; for the purpose of better illustrating the embodiments, certain features of the drawings may be omitted, enlarged or reduced, and do not represent the size of an actual product;

it will be understood by those skilled in the art that certain well-known structures in the drawings and descriptions thereof may be omitted. The technical solution of the present invention is further described below with reference to the accompanying drawings and examples.

An ant colony algorithm-based routing method applied to a vehicle-mounted self-organizing network comprises the following steps:

firstly, the method comprises the following steps: the Q learning process is implemented by the maintenance ant HELLO. In the initial stage of network operation, initializing a Q value table of each node, wherein the initial value is 0, exchanging information in the Q value table by sending a HELLO packet to a neighbor node by each node, and reading the Q value information to update the Q value table of each node after the node receives the HELLO packet.

In the Q learning model, a learning Agent and a quadruple { V, A, P, R } of the Agent are needed, wherein V represents a state set of the Agent, A represents an action selection set of the Agent, P represents a probability that the Agent selects a certain action in a certain state, and R represents an instant reward. Taking the whole VANET environment as the environment for Q learning, the corresponding relation between the elements in the VANET and the Q learning model is as follows: agent: a packet in a VANET sent from a source node to a destination node. A state set V: all vehicle nodes in the VANET. Action set A: the data packet can select the forwarding neighbor node in the VANET. P: probability of forwarding a data packet from one node to another. R: timely rewards available for forwarding packets from one node to another.

In the VANET, each vehicle node needs to maintain a Q value table, and the Q value table stores Q values corresponding to different next-hop neighbor nodes selected from the current node and used for sending data to the target node. The Q value is determined by the destination node and the next hop node, and thus the size of the Q value table is determined by the size of VANET. The update rule of the Q-value table in VANET is:

wherein μ and γ are learning rates in Q learning, respectivelyAnd discount rate, S represents current node, d represents destination node, x is a neighbor node of current node, N_xA set of neighbor nodes representing node x.

The information is updated and transmitted by the maintenance packet according to the Q value of the nodes in the network.

After the Agent in the VANET is established, Q learning is firstly carried out on each state of the Agent, and a Q value table from an initial state to a target state is obtained through the Q learning, namely, each node completes the updating work of the Q value table in parallel and cooperatively. When a node initiates a route discovery process because of needing to send data, ants start to search for a route according to a transfer rule, wherein the transfer probability is as follows:

and satisfies the following conditions:

wherein, P (d, n) and τ_(d,n)Respectively representing the transition probability and the pheromone strength of the next hop node selection node n by taking d as a destination node. S represents the node where the ant is currently located, N_sIs a set of neighbor nodes of node S, V_pRepresents the set of nodes through which the ant has passed, i ∈ N_s-V_pAnd representing the node which the node i belongs to and has not passed by the ant in the neighbor node set of the node S.

The specific implementation process is as shown in fig. 1, and specifically includes:

(1) in the initial phase of network operation, each node broadcasts a HELLO packet to neighboring nodes.

(2) After receiving the HELLO packet from the neighbor node, the current node judges whether the current node has the related information of the neighbor node which sends the HELLO packet. If the related information of the neighbor node does not exist, the neighbor node is added into the routing table of the neighbor node, and the information of the corresponding routing table entry is updated.

(3) The node reads the value in the HELLO packet and updates its Q value table according to equation (1).

(4) And after the Q value is updated, updating the transition probability information in the routing table entry of the corresponding node according to the formula (2).

(5) And after the current node finishes the updating work of the relevant information, discarding the HELLO packet.

Secondly, the method comprises the following steps: the route discovery process is realized by forward ant FANT and backward ant BANT together. When a network node wants to initiate data communication, a source node first checks whether routing information to a destination node exists, and if the routing information does not exist, a route discovery process is triggered.

When a source node initiates a route discovery process, a FANT packet is sent to a destination node, the destination node generates a response packet BANT after receiving the FANT packet and returns to the source node along a reverse route, and in the process, the BANT releases pheromones at nodes passing by, the pheromone strength of the nodes on a selected path is enhanced, and positive feedback is formed.

Defining pheromone intensity as tau_(d,n)The method is characterized in that d is used as a destination node, a next hop node is the pheromone strength of n, n is an element in a neighbor node set of a current node, and pheromone increment released by a BANT group is defined as follows:

Δτ_(d,n)＝c·Q_min (4)

where c is a constant value parameter, and its value is used as the initialization value of the pheromone, Q_minThe minimum Q value of the path traversed during this route discovery process for the FANT packet.

After the ban group completes releasing the pheromone, the pheromone table maintained by the node also needs to perform corresponding update work, i.e. overall update of the pheromone, and the rule is as follows:

τ(d,n)＝λ·τ(d,n)+(1-λ)·Δτ(d,n) (5)

in the formula, λ represents the proportion of the original pheromone in updating, and the value of λ is a set constant, and λ ∈ (0, 1).

The specific implementation process is as shown in fig. 2 and fig. 3, and specifically includes:

(1) forward ant FANTs are generated from the source node, and according to the transfer rule, the FANTs select the node with the largest transfer as the next-hop node.

(2) After receiving the FANT packet, the intermediate node firstly judges whether the same FANT packet is processed before, and if the same FANT packet is processed before, the FANT packet is directly discarded; otherwise, reading the position and speed information in the FANT packet, and calculating the link duration t by combining the position and speed information of the FANT packet_hold. Then the link duration t_holdAnd a protocol-set link duration threshold t_maxholdIn comparison, if t_holdLess than t_maxholdThen directly discard the FANT packet; otherwise, the self-related information is added into the FANT packet, and the FANT packet is updated. Finally, the intermediate node checks whether the intermediate node has the routing information to the destination node, and if so, the intermediate node forwards the FANT packet according to the information of the routing table entry; if not, the updated FANT packet is forwarded according to the forwarding rules.

(3) After the FANT reaches a destination node, firstly judging whether the same FANT packet is processed before, and if so, directly discarding the FANT packet; otherwise the link is also continued for a time t_holdAnd a protocol-set link duration threshold t_maxholdIn comparison, if t_holdLess than t_maxholdThen directly discard the FANT packet; otherwise, the destination node generates a BANT packet that inherits the relevant information in the FANT packet.

(4) The BANT packet is routed back to the source node according to the reverse route, based on the routing record of the FANT packet. In the return process, the pheromone is released on the passing node according to the formula (4), and the updating work of the pheromone is realized.

(5) After receiving the BANT packet, the intermediate node judges whether the same BANT packet is processed by the node before, and directly discards the BANT packet if the same BANT packet is processed; otherwise, according to the relevant information including the link duration of the segment on the BANT packet, updating the corresponding routing table entry and the pheromone table maintained by the node. And finally, the node forwards the BANT packet according to the reverse routing record on the BANT packet.

(6) After receiving the BANT packet from the destination node, the source node firstly judges whether the same BANT packet is processed before, and if so, directly discards the FANT packet; otherwise, updating the corresponding routing table entry and the pheromone table maintained by the node. The BANT packet is discarded after route creation is complete.

Thirdly, the method comprises the following steps: the route maintenance process is realized by detecting ant EXPLORE and maintaining ant HELLO together, and the aim is to ensure the smooth end-to-end communication. This is achieved depending on the routing duration. When the time for establishing the routing table entry exceeds the value of the routing duration in the corresponding table entry, the node sends an explicit to the destination node, which is similar to the route discovery process except for the event trigger reason of the two.

In order to make the operation of the explicit packet more efficient, the state of the neighbor node needs to be confirmed at any time, and this process is implemented by the HELLO packet. When the HELLO packet in the protocol is improved, the HELLO packet is periodically broadcast to the neighbor node, and when the node does not receive the HELLO packet sent by a certain neighbor node after a certain time, the neighbor node is automatically considered to be absent, the table information related to the node is deleted, and then an explicit packet is sent to find a new route of the destination node, and the implementation flow is shown in fig. 4.

It should be understood that the above-described embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. Other variations and modifications will be apparent to persons skilled in the art in light of the above description. And are neither required nor exhaustive of all embodiments. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the claims of the present invention.

Claims (4)

1. The ant colony algorithm-based routing method applied to the vehicle-mounted self-organizing network is characterized by being applied to the vehicle-mounted self-organizing network and specifically comprising the following steps:

step 1: the Q learning process is realized by maintaining ant HELLO; initializing a Q value table of each node at the initial stage of network operation, wherein the initial value is 0, each node exchanges information in the Q value table by sending a HELLO packet to a neighbor node, and after receiving the HELLO packet, the node reads the information of the Q value to update the Q value table of the node;

step 2: the route discovery process is realized by forward ant FANT and backward ant BANT together; when a network node needs to initiate data communication, a source node firstly checks whether routing information to a destination node exists, and if the routing information does not exist, a route discovery process is triggered;

and step 3: the route maintenance process is realized by detecting ant EXPLORE and maintaining ant HELLO together, which aims to ensure the smooth end-to-end communication; when the time established by the routing table entry exceeds the value of the routing duration time in the corresponding table entry, the node sends a detection ant EXPLORE to the destination node;

the process of the step 2 specifically comprises the following steps:

(1) generating a forward ant FANT from a source node, and then selecting a node with the largest transfer as a next-hop node;

(2) after receiving the FANT packet, the intermediate node firstly judges whether the same FANT packet is processed before, and if the same FANT packet is processed before, the FANT packet is directly discarded; otherwise, reading the position and speed information in the FANT packet, and calculating the link duration t by combining the position and speed information of the FANT packet_hold(ii) a Then the link duration t_holdAnd a set link duration threshold value t_maxholdIn comparison, if t_holdLess than t_maxholdThen directly discard the FANT packet; otherwise, adding the information of the FANT packet into the FANT packet and updating the FANT packet; finally, the intermediate node checks whether the intermediate node has the routing information to the destination node, and if so, the intermediate node forwards the FANT packet according to the information of the routing table entry; if not, continuing to forward the updated FANT packet according to the transfer rule;

(3) after FANT arrives at destination nodeFirstly, judging whether the same FANT packet is processed before, and directly discarding the FANT packet if the same FANT packet is processed; otherwise the link is also continued for a time t_holdAnd a protocol-set link duration threshold t_maxholdIn comparison, if t_holdLess than t_maxholdThen directly discard the FANT packet; otherwise, the destination node generates a BANT packet inheriting the related information in the FANT packet;

(4) the BANT grouping returns to the source node according to the reverse route according to the route record of the FANT grouping; on the way of returning, the pheromone is released on the passing node, and the updating work of the pheromone is realized;

(5) after receiving the BANT packet, the intermediate node judges whether the same BANT packet is processed by the node before, and directly discards the BANT packet if the same BANT packet is processed; otherwise, updating corresponding routing table entries and pheromone tables maintained by the nodes according to the BANT groups; finally, the node forwards the BANT packet according to the reverse routing record on the BANT packet;

(6) after receiving the BANT packet from the destination node, the source node firstly judges whether the same BANT packet is processed before, and if so, directly discards the FANT packet; otherwise, updating the corresponding routing table entry and the pheromone table maintained by the node; discarding the BANT packet after the route creation is completed;

the process of the step 3 specifically comprises the following steps:

in order to make the operation of the EXPLORE packet more effective, the state of the neighbor node needs to be confirmed at any time, and the process is realized by the HELLO packet; when the HELLO packet in the protocol is improved, the HELLO packet is periodically broadcast to the neighbor nodes, and when the HELLO packet sent by a certain neighbor node is not received by the node for a certain time, the neighbor nodes are automatically considered to be absent, the table information related to the node is deleted, and then an EXPLORE packet is sent out to find out a new route of the destination node.

2. The method according to claim 1, wherein the process of step 1 is specifically:

in the initial stage of network operation, each node broadcasts a HELLO packet to a neighbor node; after receiving the HELLO packet from the neighbor node, the current node judges whether the current node has the information of the neighbor node which sends the HELLO packet; if the information of the neighbor node does not exist, adding the neighbor node into the routing table of the neighbor node, and updating the information of the corresponding routing table item; reading the value in the HELLO packet by the node, and updating the Q value table of the node; after the Q value is updated, updating the transition probability information in the routing table entry of the corresponding node; and after the current node finishes the updating work of the information, discarding the HELLO packet.

3. The method of claim 2, wherein the Q learning model adopted only in the Q learning process needs to have a learning subject Agent and a quadruple { V, A, P, R } of the Agent, wherein V represents a state set of the Agent, A represents an action selection set of the Agent, P represents a probability that the Agent selects a certain action in a certain state, and R represents an instant reward;

taking the whole VANET environment as the Q learning environment, the corresponding relation between the elements in the VANET and the Q learning model is as follows:

the Agent corresponds to a data packet sent from a source node to a destination node in the VANET;

the state set V corresponds to all vehicle nodes in the VANET;

the action set A corresponds to a neighbor node which can select to forward in the VANET for the data packet;

p corresponds to the probability that a packet is forwarded from one node to another;

r timely rewards available for forwarding corresponding data packets from one node to another;

in the VANET, each vehicle node needs to maintain a Q value table, wherein the Q value table stores Q values corresponding to different next-hop neighbor nodes selected from the current node and used for sending data to a target node; the Q value is determined by the destination node and the next hop node, so that the size of the Q value table is determined by the scale of the VANET;

the update rule of the Q-value table in VANET is:

wherein mu and gamma are respectively the learning rate and discount rate in Q learning, S represents the current node, d represents the destination node, x is a neighbor node of the current node, N_xA set of neighbor nodes representing node x;

the information is updated and transmitted by the Q value of the nodes in the network through maintenance packets;

after the Agent in the VANET is established, Q learning is firstly carried out on each state of the Agent, a Q value table from an initial state to a target state is obtained through the Q learning, namely, each node completes the updating work of the Q value table in parallel and in cooperation; when a node initiates a route discovery process because of needing to send data, ants start to search for a route according to a transfer rule, wherein the transfer probability is as follows:

and satisfies the following conditions:

wherein, P (d, n) and τ_(d,n)Respectively representing the transition probability and the pheromone strength of the next hop node selection node n by taking d as a destination node; s represents the node where the ant is currently located, N_sIs a set of neighbor nodes of node S, V_pRepresents the set of nodes through which the ant has passed, i ∈ N_s-V_pAnd representing the node which the node i belongs to and has not passed by the ant in the neighbor node set of the node S.

4. The method of claim 1, wherein: when a source node initiates a route discovery process, a FANT packet is sent to a destination node, the destination node generates a response packet BANT after receiving the FANT packet and returns to the source node along a reverse route, and in the process, the BANT releases a certain amount of pheromones at nodes passing by, the pheromone strength of the nodes on a selected path is enhanced, and positive feedback is formed;

defining pheromone intensity as tau_(d,n)The method is characterized in that d is used as a destination node, a next hop node is the pheromone strength of n, n is an element in a neighbor node set of a current node, and pheromone increment released by a BANT group is defined as follows:

Δτ_(d,n)＝c·Q_min (4)

where c is a constant value parameter, and its value is used as the initialization value of the pheromone, Q_minMinimum Q value of the path traversed during this route discovery process for the FANT packet;

after the ban group completes releasing the pheromone, the pheromone table maintained by the node also needs to perform corresponding update work, i.e. overall update of the pheromone, and the rule is as follows:

τ(d,n)＝λ·τ(d,n)+(1-λ)·Δτ(d,n) (5)

in the formula, λ represents the proportion of the original pheromone in updating, and the value of λ is a set constant, and λ ∈ (0, 1).

Images